Steel sheet surface flatness serves as a critical quality indicator in modern industrial manufacturing, where its accurate measurement is essential for structural integrity assessment. This paper presents a vision-based measurement system using fringe projection for efficient and reliable flatness inspection. Through Fourier spectrum correction, the system extracts the spatial frequency of fringe patterns projected onto the steel surface and establishes a direct quantitative model between fringe frequency and surface flatness, enabling rapid, precise, and non-contact measurement. Comprehensive simulations and experiments were conducted to analyze key factors influencing system performance, confirming the feasibility and reliability of the proposed approach. Comparative tests against conventional stereoscopic vision demonstrate that the proposed system offers superior robustness to highly reflective surfaces, reduced system complexity, and higher operational efficiency. The core innovation lies in integrating spectral correction with a direct analytical model, which maintains high measurement accuracy while effectively suppressing specular reflections and significantly simplifying the system structure, thereby lowering costs. This provides a practical and innovative solution for online, rapid, and non-contact flatness inspection of steel sheets in industrial settings.

We present an optical non-contact torque sensor for quasi-static torque measurement that complies with the mechanical definition of torque by measuring the relative twist between two shaft cross-sections via a dual-path, single-source interferometric readout. Frequency-division multiplexing enables simultaneous acquisition from both sections using a single spectrometer while avoiding crosstalk. To counter the reduced signal power and spectral leakage introduced by multiplexing, we develop a synchrosqueezing-based Hanning-window energy centrobaric method (SST-HnWECM) that enhances weak spectral components and improves frequency estimation. A physics-based model with complete mathematical derivation is provided, and the approach is implemented on an OCDM platform with laser-engraved circumferential scales that map linear displacement to twist angle. Experiments demonstrate a non-linear error of 0.05 % and a repeatability error of 1.30 %, while retaining the practical advantages of optical readout-contactless operation and vibration robustness-for micro-torque measurement. These results indicate that the proposed dual-section scheme and SST-HnWECM processing provide a viable and accurate solution for optical torque sensing.

As industrialization advances and the global push for carbon neutrality intensifies, enhancing the efficiency of mechanical equipment has become essential. Accurate measurement of rotational torque plays a crucial role in monitoring efficiency and ensuring optimal performance. This article presents a novel, noncontact measurement technique based on optical coherent displacement, integrating dual-optical-detector probes for simultaneous rotational torque and speed measurement. A sensing model that links optical coherent signals to changes in rotational torque and rotational speed is established. The experiment demonstrates high accuracy of the system, with rotational speed error ranging from 0.25% to 0.67%, and rotational torque indication error between 0.03% and 2.25%. Furthermore, the experiments proved that the repeatability error is less than 1%, the hysteresis error is less than 1.6%, and the linearity error is in the range of 0.24% similar to 0.57% for rotational torque measurement. The research further evaluates the influence of rotational speed on rotational torque measurement accuracy, revealing minimal impact at higher speeds. The findings suggest that the proposed method offers significant potential for precision measurement in rotating machinery, enabling the simultaneous measurement of both torque and rotational speed. This capability has important implications for improving system efficiency and supporting sustainable industrial practices.

Bistable composite tape-spring (CTS) structures are favoured for space deployable applications owing to their self-latching ability; however, their actuation often relies on heavy, complex mechanical systems. To overcome this, we have established a smart driving strategy using integrated smart materials to enable efficient, remote morphing, where a bilayered intelligent structure comprises an active smart-material layer and a passive, antisymmetric CTS layer. An analytical model was developed to predict structural bistability and map the shape-changing process via minimum strain energy paths. This framework characterizes how stimulus-induced strain in the active layer applies an equivalent bending moment to drive the CTS. Key design criteria are quantified: the CTS must operate within specific moment boundaries and the active layer must not exceed a critical thickness to ensure bistable snapping. Furthermore, the required actuation strain is shown to decrease predictably with increasing active layer thickness, with analytical predictions showing a mean error of 6.0% against numerical results. These results provide definitive guidelines for designing smart, bistable morphing structures. Consequently, this work contributes a foundational model and clear material-geometric criteria to facilitate the development of lightweight, low-complexity, and energy-efficient smart deployable structures for aerospace applications.

Fiber-reinforced composite metamaterials (FRCMs) represent a class of emerging structural absorbing materials for load-bearing components in aerial combat aircraft, combining wave absorption capability with bearing property to enhance survivability. FRCMs are typically fabricated by thermal curing, where inevitable temperature inhomogeneities in large components intensify thermal mismatch and asynchronous curing between metamaterials and composites, resulting in defects and deformation that compromise load-bearing and stealth performance. Therefore, this study optimizes the design of FRCMs to achieve broadband stealth performance (4–18 GHz) with enhanced absorption at 2.45 GHz. By adopting microwave curing instead of conventional thermal curing, uniform isothermal heating is realized, enabling efficient and high-quality fabrication. The results demonstrate that the designed square-loop resonators (SLRs) achieve efficient microwave absorption through electromagnetic resonance at 2.45 GHz, enabling uniform heating of FRCMs, with the in-plane temperature difference controlled within 5℃. Furthermore, multilayer split-ring resonators (SRRs) are constructed, which exploit destructive interference between the layers to achieve broadband stealth functionality across the 4.6–18.0 GHz range, with an arithmetic average absorbance exceeding 90

Thermal barrier coatings (TBCs) are essential for thermal insulation and oxidation protection in turbine engines, yet their failure due to micro-cracking and spallation remains a critical challenge. This failure is primarily driven by complex internal stresses arising from thermal expansion mismatch, dynamic growth of the thermally grown oxide (TGO) layer, and inherent microstructural defects. While prior research has offered insight, prevailing two-dimensional models often overlook the dynamic nature of TGO evolution and the three-dimensional complexity of pore networks, limiting a holistic understanding of stress mechanisms. Here, we have developed a three-dimensional finite element model to investigate stress evolution under isothermal cycling, incorporating dynamic TGO growth, material creep, diverse 3D pore morphologies, and pore-interface roughness interactions. Results show that TGO growth progressively elevates interfacial stresses, while creep provides critical relaxation. The presence of pores significantly amplifies stress magnitudes, with morphology and orientation critically influencing local concentrations. Furthermore, a strong non-linear coupling exists between interface roughness and pores, where their combined effect peaks at a specific roughness ratio. These findings elucidate the complex, multi-factor failure mechanisms in TBCs and provide guidelines for designing more durable coating systems.

Thermal barrier coatings (TBCs) are widely employed in aviation engines due to their high thermal insulation properties. However, the formation of thermally grown oxide (TGO) during thermal cycling promotes micro-crack propagation, ultimately leading to the premature failure of TBCs. Nevertheless, non-destructive detection of weak defects remains a significant challenge for TBCs. Here, a sparse deconvolution method is introduced to measure the coating thickness and enhance the axial resolution of terahertz signals, thereby overcoming the challenge of detecting weak defects induced by overlapped terahertz echoes. More importantly, a total variation fusion imaging approach is proposed for detecting TBCs defects under a time sequence of thermal cycling, which reveals the generation and expansion behaviors of weak defects. The experimental results demonstrate that the measurement resolution of terahertz is approximately 10 μm in the axial direction, and reveal the propagation behaviors of cracks and delamination under thermal cycling. This research provides a new non-destructive evaluation method for TBCs and offers new insights into the reliability assessment in aerospace applications.

ABSTRACT Radar stealth technology is pivotal for next‐generation air dominance, necessitating the integration of shaping and material stealth to minimize radar cross‐section. Radar absorbing materials (RAMs), which dissipate incident electromagnetic waves as heat, are crucial for overcoming the aerodynamic compromises of shaping stealth. The performance of conventional RAMs is largely constrained by their thickness and filler morphology. The emergence of 2D materials presents a revolutionary opportunity for material stealth due to their unique lamellar structures, tunable electronic properties, low density, and high specific surface area. These characteristics provide more controllable pathways for enhancing absorption performance. This review focuses on recent advances in 2D materials for RAMs, highlighting absorption enhancement and modulation methods. It begins with an overview of electromagnetic wave loss mechanisms and the role of 2D materials. Subsequently, various enhancement strategies are discussed, including the design and fabrication of porous structures, heterogeneous interfaces, printed metamaterials, and tunable metasurfaces. The review concludes by outlining existing challenges and offering perspectives on the future development of 2D materials empowered RAMs.

This paper proposes a decoupling method for column-type six-axis force/torque sensors based on a physics-informed neural network (PINN). The method incorporates the deformation behavior of the sensor elastic body, the sensing mechanism of strain gauges, and the Wheatstone bridge principle as physical prior constraints, establishing a hybrid model composed of a physics-informed decoupling layer and a deep error compensation network that operate synergistically. The decoupling layer suppresses cross-axis coupling by enforcing these constraints, while the compensation network compensates for residual nonlinear errors. The method was compared with LS and BPNN under varying sample sizes, calibration point numbers, and external loading points. With 256 repeated measurements per calibration point, the RMSPE is 0.97% and coupling errors are within 3.0%; with only four repetitions, the RMSPE decreases to 0.87% and coupling errors below 2.0%. When calibration points are reduced to one, indication and coupling errors in most directions remain below 5%, while LS and BPNN errors escalate dramatically. At external loading points, the PINN achieves an overall RMSPE of 0.81% and coupling errors below 2.0%, showing excellent generalization. Physical prior constraints effectively reduce dependence on data volume and point density, providing a promising solution for multi-axis force sensor decoupling in data-scarce scenarios.

Accurate and stable rotational speed measurement is essential for motion control and condition monitoring. This paper presents a vision-based method utilizing inclined sinusoidal fringes, where rotational angles are encoded as phase shifts and decoded via a dual phasecorrelation algorithm. Unlike traditional fringe processing methods that directly rely on gray intensity values and are thus more susceptible to noise and spectral interference, this approach enhances robustness. The first phase correlation step extracts sinusoidal Pearson correlation coefficient curves from the fringe images, while the second correlation calculates inter-frame phase shifts for instantaneous rotational speed estimation. A mathematical imaging model is developed, followed by simulations to evaluate the impact of signal characteristics, noise, and other factors. The proposed method was validated for high accuracy and stability through repeatability tests against a high-precision encoder. The average relative errors obtained from repeated experiments were 0.8796% at 30 rpm, 0.0911% at 600 rpm, and 0.0479% at 1200 rpm. Moreover, at low speeds, the proposed method delivered noticeably smoother speed profiles than the encoder. Furthermore, experiments with dynamic rotational speed on a rotor test platform show that it outperforms conventional encoders, especially at low speeds and during rapid fluctuations. The high sampling rate of the proposed method ensures smooth speed tracking, making it well-suited for dynamic analysis, condition monitoring, and fault diagnosis in rotating systems.

The fourth industrial revolution (Industry 4.0) has transformed manufacturing by integrating machines, materials, and human inputs through advanced operational and information technologies. This transformation has likewise reshaped nondestructive evaluation (NDE), giving rise to NDE 4.0. This new paradigm is enabled by key technologies such as Artificial Intelligence (AI), digital twins (DTs), automation, smart/cognitive sensors, and the Industrial Internet of Things (IIoT). These technologies enhance quality assurance, streamline workflows, enable real-time data-driven decision-making, and facilitate intelligent asset management. While NDE 4.0 is continuously transforming traditional practices, critical sectors such as aerospace, energy, and biomedical imaging still rely on conventional NDE methods, which struggle with scalability, data silos, and insufficient real-time analytics. To address these gaps, this paper offers a comprehensive strategic review and a forward-looking roadmap for advancing NDE 4.0 toward NDE 5.0. It synthesizes state-of-the-art developments while proposing a data-centric framework that integrates evolving DT technologies (from static to dynamic multi-tiered systems), autonomous inspection platforms, and hybrid data-fusion methods. Practical deployment challenges such as workforce skill gaps, cybersecurity, sustainability, and regulatory standardization, are examined alongside transformative emerging technologies such as terahertz sensing, quantum techniques, robotics and drones, augmented and virtual reality-assisted frameworks, blockchain, and big-data analytics. Recognizing the importance of ethical AI and open-science principles, this study advocates for a cohesive, globally standardized approach to accelerate NDE 4.0 adoption. Looking ahead, this work envisions NDE 5.0 as a synergistic human-machine ecosystem featuring self-learning, autonomous inspection systems that deliver enhanced reliability, safety, and efficiency across industries.

Terahertz (THz) metamaterial absorbers are essential for suppressing electromagnetic interference, yet practical devices must also combine visible transparency and tunability to meet emerging needs in stealth and aerospace technologies. Current approaches struggle, since most active materials enabling tunability are optically opaque. In this work, we present a transparent, dynamically tunable THz absorber (TD-TMA) based on a reversible electrochemical approach. The TD-TMA consists of a micro-cross-patterned top ITO, a planar bottom ITO, and a sandwiched Ag+-based electrolyte gel, ensuring optical transparency. Voltage polarity switching induces reversible Ag+ deposition at the cross edges, altering the resonant unit dimensions and tuning the absorptivity frequency. Experiments indicate that the TD-TMA demonstrates perfect-absorption at 0.547 THz with a tunable range from 0.547 THz to 0.562 THz, with broadband-adjustable absorptivity from 1.2 to 2.0 THz. Notably, the device retains high visible-light-transparency (81.8%) during modulation. This electrochemical tuning strategy delivers a new design route for transparent THz absorbers - promising for applications ranging from wearable devices to communication systems.

High-scattering ceramics, such as thermal barrier coatings (TBCs), are primary targets for mid-infrared optical coherence tomography (MIR-OCT). Compared with conventional near-infrared OCT, MIR-OCT systems leverage long-wavelength light sources to suppress optical scattering significantly, achieving substantial breakthroughs in both penetration depth and signal-to-noise ratio (SNR). However, current optimizations of the system’s anti-scattering performance primarily focus on a single approach: increasing the wavelength to reduce scattering. This study demonstrates that, in addition to the inherent advantages of longer wavelengths, the anti-scattering performance of MIR‑OCT can be further enhanced by discriminating the spatial and polarization characteristics of backscattered light. We first determined the scattering and absorption coefficients of TBCs in the mid-infrared range using a four-flux Kubelka-Munk (4F-KM) method. These parameters were integrated into the polarimetric Monte Carlo multi-layer (PMCML) model to study the relationship between the scattering frequency of backscattered photons and their spatial distribution and polarization evolution. Based on this analysis, two anti-scattering strategies are proposed: spatial filtering and polarization-based filtering. Simulation results demonstrate that both approaches effectively suppress multiply scattered photons. Experiments further confirm that spatial filtering, as one of the two strategies, substantially reduces scattering-induced background noise, validating its effectiveness in improving MIR‑OCT system performance.

Large-range displacement measurement is essential for mechanical structural health monitoring, yet existing methods often struggle to balance measurement range, accuracy, robustness, and cost. To address this issue, a dynamic large-range displacement measurement method based on composite fringe vision sensing is proposed. A crossed composite fringe pattern is designed as the optical marker, and a sub-pixel strategy combining dynamic region-of-interest updating and spectral zero-padding is developed to achieve large-range, high-precision displacement measurement and stable tracking of complex two-dimensional motion trajectories. Simulation results show that the method maintains stable performance under illumination coefficient variations from 0.5 to 1.6, SNR levels from 0 to 40 dB, and different defocus conditions. In addition, accurate displacement measurement is still achieved at a relatively low resolution of 175 × 175 pixels. Experimental results demonstrate a minimum detectable step displacement of 15 μm, while repeatability remains within 15 μm over one-dimensional displacement ranges from 75 mm to 600 mm. The proposed method provides an efficient, economical, and reliable solution for displacement measurement in mechanical structural health monitoring.

Chip packaging defect detection is critical for ensuring system operational stability. Addressing the low efficiency and poor generalization of traditional manual visual inspection and early machine vision detection, and considering the challenges posed by the small size and subtle characteristics of common chip packaging defects, this paper proposes a CPD-YOLO algorithm based on YOLOv8n. Firstly, an overparameterized feature extraction module (C2f-DO) is designed to enhance the model's feature extraction capability; secondly, an attention feature orthogonal fusion module is introduced to orthogonally integrate global and local semantic information across multiple scales, thereby strengthening the discriminative representation of small defects. Subsequently, a Haar wavelet downsampling and upsampling module is constructed to enhance the resolvability of faint defect signals in the frequency domain. Finally, we introduce InterpIoU regression loss to improve the stability and accuracy of bounding box regression for small-scale targets. On our self-built chip packaging defect dataset, CPD-YOLO achieves a mAP of 96.6%, representing a 3.9% improvement over baseline models. Concurrently, model parameters and computational complexity are reduced by 26.6% and 44.4%, respectively. Furthermore, experiments on other datasets demonstrate that CPD-YOLO outperforms alternative detection methods, validating the superiority and practicality of the proposed approach.

Plasmonic resonators represent an emerging toolkit in biosensing by focusing electromagnetic waves within strongly enhanced electric field hotspots to facilitate light-matter interaction. However, the limited efficiency of field enhancement and the low spatial overlap between analytes and plasmonic hotspots impose fundamental limitations on achieving high-performance molecular assays. In this work, a terahertz (THz) biosensor for specific detection that allows molecules in solution to readily access predefined strong hotspot regions is presented. We construct heterogeneous nanogap structures by introducing a 20 nm-thick SiO2 nanofilm to efficiently localize THz waves into the gaps providing over 300-fold field enhancement and optimal spatial matching with molecules. The sulfhydryl compounds are utilized to form Au-S bonds to cover large areas of the gold surface, preventing the aggregation of target molecules in the non-hotspot regions. Furthermore, the exposed silica surfaces within the nanogaps are functionalized with a biotin monolayer via silane coupling, enabling targeted active capture trace amounts of free streptavidin molecules in solution for specific detection at low concentrations down to 150 fM. This approach provides a valuable toolkit for ultra-sensitive biochemical detection and opens new avenues for the development of photonic devices in multifunctional, real-time, and on-chip sensing.

Tunable absorbers hold significant application prospects in fields such as detection, communication, and electromagnetic pollution mitigation. However, conventional designs often struggle to achieve high efficiency absorption during dynamic tuning due to the mismatch between radiative leakage and intrinsic dissipation. To address this, this paper proposes a liquid crystal-integrated jigsaw-like terahertz metamaterial absorber and reveals a dual-resonance mode critical coupling mechanism based on temporal coupled mode theory. The proposed structure consists of two orthogonally arranged resonant units, each supporting a plasma resonance mode. For both modes, the radiative leakage channels are simultaneously reconfigured through a shared geometric coupling parameter, enabling precise control over their coupling states. At a specific parameter configuration, both resonant modes simultaneously satisfy the critical coupling condition—where radiative loss equals intrinsic absorption loss—thereby achieving near perfect absorption at two frequency bands. Furthermore, by applying voltage to modulate the liquid crystal, the resonance frequencies are rendered tunable. The results demonstrate that during tuning, the absorption amplitudes at both resonance frequencies remain above 99%, with relative tuning ranges of 6.22% and 5.13% for the two absorption peaks, respectively. This study offers a novel and effective design strategy for developing high performance, tunable terahertz functional devices.

Space deployable structures are developing towards large- or even mega-scale level to meet practical application requirements for future deep space explorations. A bistable composite tape-spring (CTS) has attracted great attention since it is superior in terms of lightweight, low cost, and high storage-to-package ratio. Space applications of the thin-walled CTS structures are also getting longer, which tend to become distorted since they are usually produced with antisymmetric unidirectional (UD) composite layups; whilst the acting mechanics is still unrevealed, leading to significant difficulties in deployable stability control and structural integrity design. Here, we present a detailed study on exploring the antisymmetric bistable mechanics of the CTS structure. This is achieved by investigating the antisymmetric UD layup-induced thermal residual stress (TRS), as well as the effects on stress, strain and curvature of the produced CTS. A strain energy analytical model is developed by extending the well-established two-parameter inextentional model by integrating the TRS effects, so that the strain energy evolution is disclosed to reveal the underlying mechanics; the minimum strain energy paths are then plotted to define the theoretical boundaries on structural bistability, which show good agreement with experimental observations and finite element analysis. The refined analytical model is found to be accurate in predicting the TRS-induced structural distortion of a CTS with arbitrary antisymmetric layups; the antisymmetric bistable mechanisms are then discussed to expose further insight. These findings extend the classical bistability model to include curing-induced thermal effects, offering new insight into the antisymmetric mechanics of the CTS structure. This work is expected to enrich the design and manufacturing of large-scale deployable composites for aerospace applications.

Multifunctional terahertz absorbers require switchable narrowband and broadband responses. In the low-terahertz region, conventional planar multi-resonant structures suffer from strong near-field coupling, which hinders the simultaneous realization of broadband and high-Q narrowband absorption and often leads to spectral distortion. Here, we propose a vertically stacked terahertz metamaterial absorber based on vanadium dioxide (VO2) and silicon resonators, where interlayer electromagnetic coupling is actively controlled by the VO2 phase transition. In the insulating state, the VO2 layer becomes a low-loss transparent medium, enabling coherent coupling between the guided-mode resonance (GMR) of the top silicon grating and the asymmetric Fabry–Pérot (F-P) cavity mode in the bottom layer. Under phase matching, critical coupling yields ultra-narrowband absorption >99.7% at 0.691 and 0.782 THz with Q-factors of 265.4 and 217.2. In the metallic state, strong Ohmic losses trigger resonance quenching that suppresses the top-layer narrowband resonance, and the device becomes dominated by a metallic VO2-based metal–insulator–metal (MIM) configuration, achieving broadband absorption >90% across 0.326–0.717 THz. Our vertical stacking and resonance suppression strategies physically eliminate undesired mode coupling in the low-terahertz band, enabling crosstalk-free switching between narrowband and broadband absorption, thus offering a new paradigm for high-performance multifunctional terahertz devices.

Intelligent mechanical fault diagnosis has achieved remarkable advancement in recent years, yet it remains dominated by contact sensors whose installation constraints, load effect, invasive measurement, and single-point acquisition severely restrict applicability in complex environments. To overcome these limitations, this paper proposed a framework that synergizes a fringe projection vision system with a Dual-Weight Enhanced Prototype Network (DWEPN) for non-contact bearing fault diagnosis. The main contributions of this work were summarized as follows: 1) A fringe projection system is employed to acquire radial shaft vibration in a non-contact manner, eliminating sensor installation and load effects while enabling multi-point, material-independent, and long-distance measurement. 2) Shaft center orbit images reconstructed from radial vibration are used as diagnostic inputs, providing intuitive geometric representations of defect-induced deviations while preserving richer fault information by shortening vibration transmission paths. 3) To address the limitations of conventional prototype networks, a Dual-Weight Enhanced Prototype Network is proposed, incorporating adaptive weighting during both prototype construction and metric evaluation. Specifically, an Adaptive Prototype Weighting Network assigns sample-specific weights to mitigate prototype shift caused by heterogeneous support sample quality, instead of generic prototype refinement. Meanwhile, a Query Channel Weight Generator selectively enhances fault-sensitive channels in query features prior to distance computation, explicitly improving query feature discrimination rather than applying a unified attention mechanism across all learning stages. Finally, experimental results demonstrated that the proposed method achieved superior diagnostic performance across different operational conditions compared to existing meta-learning methods, providing a non-contact solution for industrial equipment intelligent maintenance that combines high precision with strong generalization capability.